Merge branch 'next' into guile.

Passes distcheck.

* next: (32 commits)
  volk: Fix for popcnt's 64/32-bit issues.
  Using a copy of config.guess and config.sub instead of sym links.
  Including time header to qa files.
  Changed python env variable to more globally usable version.
  gr_uhd: Quick fix for make distcheck failures if UHD is not installed.
  PFB resampler: fix it this way to avoid the signed/unsigned warning.
  PFB resampler: fixes bug where filter could be looking past the number of inputs.
  resampler PFB: Sets relative rate when rate is changed.
  Create method to set rate on pfb_arb_resamp after it has been created. Allow it to be called from GRC.
  volk: fix for running 32-bit OS on 64-bit processor. System is correctly identified as 32-bit and compiles with the correct flags.
  volk: changing the path variables again. This works on my various systems tested. Using abs_ path names failed on Ubuntu 8.04 32-bit.
  volk: May be a hack, but it was required for my 32-bit Fedora 13 to work.
  volk: Removing unnecessary shell script; last commit takes care if its functions.
  volk: Fixing build system to handle making volk_mktables, volk_tables.h, and volk_config.h instead of a standalone shell script.
  volk: readding 16sc_magnitude_32f_sse with fix for SSE hadd_ps error.
  volk: Adding a few more generic-only test cases.
  volk: adding generic QA test for 16sc_magnitude_32f.
  volk: modified the configure scripts to output which architectures it will be building based on the configure tests.
  uhd: update notes in grc blocks for addressing scheme
  volk: Removing erroneous SSE function that actually usese an SSE3 intrin (mm_hadd_ps).
  ...

1 files changed, 111 insertions, 0 deletions

diff --git a/volk/lib/qa_16sc_magnitude_32f_aligned16.cc b/volk/lib/qa_16sc_magnitude_32f_aligned16.cc
new file mode 100644
index 000000000..1ebe644c5
--- /dev/null
+++ b/volk/lib/qa_16sc_magnitude_32f_aligned16.cc
@@ -0,0 +1,111 @@
+#include <volk/volk.h>
+#include <qa_16sc_magnitude_32f_aligned16.h>
+#include <volk/volk_16sc_magnitude_32f_aligned16.h>
+#include <cstdlib>
+#include <ctime>
+
+//test for sse
+
+#ifndef LV_HAVE_SSE3
+
+void qa_16sc_magnitude_32f_aligned16::t1() {
+  clock_t start, end;
+  double total;
+  const int vlen = 3201;
+  const int ITERS = 10000;
+  std::complex<int16_t> input0[vlen] __attribute__ ((aligned (16)));
+  
+  float output_generic[vlen] __attribute__ ((aligned (16)));
+  float output_known[vlen] __attribute__ ((aligned (16)));
+
+  int16_t* inputLoad = (int16_t*)input0;
+  for(int i = 0; i < 2*vlen; ++i) {   
+    inputLoad[i] = (int16_t)(rand() - (RAND_MAX/2));
+  }
+  printf("16sc_magnitude_32f_aligned\n");
+
+  float scale = 32768.0;
+  for(int i = 0; i < vlen; ++i) {   
+    float re = (float)(input0[i].real())/scale;
+    float im = (float)(input0[i].imag())/scale;
+    output_known[i] = sqrt(re*re + im*im);
+  }
+
+  start = clock();
+  for(int count = 0; count < ITERS; ++count) {
+    volk_16sc_magnitude_32f_aligned16_manual(output_generic, input0, scale, vlen, "generic");
+  }
+  end = clock();
+  total = (double)(end-start)/(double)CLOCKS_PER_SEC;
+  printf("generic_time: %f\n", total);
+
+  /*
+  for(int i = 0; i < 100; ++i) {
+    printf("inputs: %d + j%d\n", input0[i].real(), input0[i].imag());
+    printf("generic... %f == %f\n", output_generic[i], output_known[i]);
+  }
+  */
+  
+  for(int i = 0; i < vlen; ++i) {
+    //printf("%d...%d\n", output0[i], output01[i]);
+    CPPUNIT_ASSERT_DOUBLES_EQUAL(output_generic[i], output_known[i], fabs(output_generic[i])*1e-4);
+  }
+}
+
+#else
+
+void qa_16sc_magnitude_32f_aligned16::t1() {
+  
+  volk_environment_init();
+  clock_t start, end;
+  double total;
+  const int vlen = 3201;
+  const int ITERS = 100000;
+  std::complex<int16_t> input0[vlen] __attribute__ ((aligned (16)));
+  
+  float output_generic[vlen] __attribute__ ((aligned (16)));
+  float output_sse[vlen] __attribute__ ((aligned (16)));
+  float output_sse3[vlen] __attribute__ ((aligned (16)));
+
+  int16_t* inputLoad = (int16_t*)input0;
+  for(int i = 0; i < 2*vlen; ++i) {   
+    inputLoad[i] = (int16_t)(((float) (rand() - (RAND_MAX/2))) / static_cast<float>((RAND_MAX/2)));
+  }
+  printf("16sc_magnitude_32f_aligned\n");
+
+  start = clock();
+  for(int count = 0; count < ITERS; ++count) {
+    volk_16sc_magnitude_32f_aligned16_manual(output_generic, input0, 32768.0, vlen, "generic");
+  }
+  end = clock();
+  total = (double)(end-start)/(double)CLOCKS_PER_SEC;
+  printf("generic_time: %f\n", total);
+  start = clock();
+  for(int count = 0; count < ITERS; ++count) {
+    volk_16sc_magnitude_32f_aligned16_manual(output_sse, input0, 32768.0, vlen, "sse");
+  }
+  end = clock();
+  total = (double)(end-start)/(double)CLOCKS_PER_SEC;
+  printf("sse_time: %f\n", total);
+
+  start = clock();
+  for(int count = 0; count < ITERS; ++count) {
+    volk_16sc_magnitude_32f_aligned16_manual(output_sse3, input0, 32768.0, vlen, "sse3");
+  }
+  end = clock();
+  total = (double)(end-start)/(double)CLOCKS_PER_SEC;
+  printf("sse3_time: %f\n", total);
+
+  for(int i = 0; i < 1; ++i) {
+    //printf("inputs: %d, %d\n", input0[i*2], input0[i*2 + 1]);
+    //printf("generic... %d, ssse3... %d\n", output0[i], output1[i]);
+  }
+  
+  for(int i = 0; i < vlen; ++i) {
+    //printf("%d...%d\n", output0[i], output01[i]);
+    CPPUNIT_ASSERT_DOUBLES_EQUAL(output_generic[i], output_sse[i], fabs(output_generic[i])*1e-4);
+    CPPUNIT_ASSERT_DOUBLES_EQUAL(output_generic[i], output_sse3[i], fabs(output_generic[i])*1e-4);
+  }
+}
+
+#endif